While data provided by Helium 10 is derived from Amazon and strives to be as accurate as possible, the numbers may not always be 100% current as they are only estimations. Additionally, some products that possess more variability such as clothing with multiple sizes, colors, shapes, etc. are more difficult to estimate. Other reasons for mismatched numbers can include delayed or insufficient data, model imperfection or simply a bug.
Some of the data is 'actual' - like keyword search volume, BSR (current and historic), review count and avg. review, number of sellers. The sales (and revenue, which is derived from sales) are estimate. The accuracy (or more precisely veracity) of estimation depends on the quality of the model and the quality of data that is used with the model. We continue working on improving both. We run tests against main competitors and based on the data points that we have available our estimation is comparable or better on most categories and for BSR, range for which we optimize our model, which is 1,000 - 100,000 BSR. This range changes with categories, but that is fair approximation. The model will perform better on products that sell at least few units per day. It will not work great on products that sell 0-2 units/day.
The tools are not designed to produce 100% accuracy even on the data points that we do have available. Again, our baseline is competition and based on what we are seeing in most cases we tested, we are doing as well or better.